OBJECTIVES: To estimate the economic impact of (1) treating pregnant women who are human immunodeficiency virus (HIV)-positive with zidovudine and (2) voluntary screening programs for pregnant women for HIV infection and offering treatment with zidovudine to those found to be HIV-positive. MAIN OUTCOME MEASURES: Number of cases of pediatric HIV infection and costs of screening, zidovudine treatment, and pediatric HIV infection treatment. DESIGN: Health care costs associated with treatment of HIV-positive pregnant women and their newborns are estimated as the costs of zidovudine and its administration and the reduction in costs of treating pediatric HIV infection. The lifetime costs of pediatric HIV infection are derived from the published literature. Estimates of the reduction in maternal-to-fetal transmission rates are taken from the AIDS [acquired immunodeficiency syndrome] Clinical Trials Group (ACTG) Protocol 076. Costs of a voluntary screening program include costs of screening tests and counseling. Sensitivity and threshold analyses are performed to determine the impact of changes in input parameter values including zidovudine treatment costs, efficacy of treatment, costs of pediatric HIV infection, prevalence of HIV infection in pregnant women, screening test sensitivity and specificity, and pregnancy termination rates on the results. RESULTS: Assuming transmission rates are reduced from 25.5% to 8.3% as found in the ACTG 076 trial, treatment costs of $104,502 for 100 HIV-positive pregnant women and their newborns are offset by the reduction of $1,701,333 associated with fewer cases of pediatric HIV infection for a net savings of $1,596,831. The sensitivity and threshold analyses show that overall cost savings from treatment of HIV-positive pregnant women and their newborns are achieved for a wide range of possible maternal treatment costs, efficacy rates, and lifetime pediatric HIV treatment costs. In the base-case analysis for the voluntary screening program, overall cost savings are seen when HIV prevalence rate among pregnant women is greater than 4.6 per 1000. However, this threshold prevalence rate is sensitive to changes In parameter value-especially pediatric HIV treatment costs, counselling costs, efficacy of treatment, and years of additional HIV treatment for the pregnant women. CONCLUSIONS: Offering zidovudine treatment to pregnant women known to be HIV-positive will decrease the number of cases of pediatric HIV infection and reduce health care costs. Voluntary screening programs for pregnant women will further decrease the number of cases of pediatric HIV infection. The effect of a screening program on health care costs varies according to HIV prevalence and the costs associated with the screening program.